Hello,
In my project I have an object that moves in x0y plane from one point to another ( A to a A' -forth and lift) and then maybe to another close point ( the average distance between points is around 20-30 cm). I need to know when my object reaches each point, the sensor has to be on the object, I can't have marks or magnets on those points. There is no tilt, no roll, no yaw so an IMU won't work or does it?
Thank you.
You can use an accelerometer to calculate the distance, but it will be highly inaccurate.
Why can't you use marks or magnets ?
Could you use infra-red, or doppler radar, or laser beams ?
Can you tell us more about it ?
What kind of objects ? What kind of floor ?
Perhaps someone has a good idea to solve it in a different way.
I need only to know when the object reaches A and A', not the distances, the points are fixed. The object is just a box (where i stack my sensors and controllers) attached to an axle. It moves back and forth ( up and down) from A to A' and back remaining parallel to the ground.
Measuring the ground clearance with some IR distance sensor could be an option, but with a couple of downsides -possible objects blocking the path, isn't there a more 'elegant' way? Not the radar thingy, nor optic systems. 
The sensor must be integrated in the box, it has to be done from the inside, autonomous. The arm (axle) lifts the box in the air so magnets and markings are not an option.
In other words, the 'blind' box needs to know if it's located in A or A'. A and A' always describing an fixed size hypotenuse, like in the drawing.
If an arm (axle) is used to move the box, the arm should keep track of the angle. It that an arm with stepper motors or could you add rotary position sensors on the arm, or perhaps even accelerometer/gyro sensors ?
The distance 20-30cm is something that could be measured with an ultrasonic distance sensor. Either IR or ultrasonic could be used to measure the height.
What kind of accuracy do you need ?
Perhaps a combination of sensors could be used.
I think that a laser beam that is interrupted by the box is a good way. Why can't you use something optical ?
In the end, it all comes down to the arm. That is where the location is known.
So far my thoughts, I can't come up with something else.
First things first, so thank you for your time and answers.
The axle is off-limits for my device, so I can't measure the axle tilt, etc Laser unusable because all the sensors must be in the box.
I need only to know when and how many times the box reaches each point.
IR sensors measuring the height might be the answer, but there could be a dust problem, or droplets of water disturbing the measurement.
If I'm understanding correctly, you are looking for a sensor to double check that your robot is where you think it is at certain points, correct?
You could do something with a hall effect sensor and some magnets. It will detect when the magnet is nearby. You can mount the sensor on the arm of whatever you are moving, and embed the magnets into your platform, and assuming there is not some large amount of clearance required, you should be able to detect when you get within some distance of the magnet. The sensors would only be blocked by ferromagnetic material (IE, Steel), but will ignore particles in the air, and it is non-contact. Precision will depend on your sensors and setup, but it might be worth looking into.
You seem to be ruling out all the practical options of using direct position sensing to measure where the object is. It would be worth making absolutely sure you have to rule those out, because this feels like a problem that could be made much easier with just a bit of lateral thinking. However, without having any idea of the actual constraints you're working within we can only take what you say at face value and accept that you cant attach anything to anything.
Given those constraints, the way I'd solve this is with an external optical sensor which detected the position of the box. Either a simple beam-breaking system or, if you've going to introduce constraints that prevent that, then with a camera and image processing solution.
If you want the detector to operate blind inside the box with no access to the rotating parts, could you embed a small ferrite coil at each of the target points, and have a coil in the box detecting either of the frequencies, I don't know what accuracy you want, but I have used this before and it worked. ( it was a LONG time ago, and I used the coils from the tape end detect of a Sony Betamax VCR, but I dont think you would find them any more )
The project was put on hold a couple of years ago, and it will be resumed soon.
PeterH, that's correct, I can't attach anything to anything, image processing won't work, but a simple beam-breaking system might do the trick. Anyway, I have to check the novelties in sensors first.
Thank you all for your time and ideas.
